CN105247860B - Image processing equipment and image processing method - Google Patents

Image processing equipment and image processing method Download PDF

Info

Publication number
CN105247860B
CN105247860B CN201480030688.6A CN201480030688A CN105247860B CN 105247860 B CN105247860 B CN 105247860B CN 201480030688 A CN201480030688 A CN 201480030688A CN 105247860 B CN105247860 B CN 105247860B
Authority
CN
China
Prior art keywords
parallax value
viewpoint
anaglyph
generation unit
correlation
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201480030688.6A
Other languages
Chinese (zh)
Other versions
CN105247860A (en
Inventor
井藤功久
早坂健吾
森浩典
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Sony Corp
Original Assignee
Sony Corp
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Sony Corp filed Critical Sony Corp
Publication of CN105247860A publication Critical patent/CN105247860A/en
Application granted granted Critical
Publication of CN105247860B publication Critical patent/CN105247860B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Classifications

    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
    • H04N13/20—Image signal generators
    • H04N13/204—Image signal generators using stereoscopic image cameras
    • H04N13/207—Image signal generators using stereoscopic image cameras using a single 2D image sensor
    • H04N13/225—Image signal generators using stereoscopic image cameras using a single 2D image sensor using parallax barriers
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/97—Determining parameters from multiple pictures
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
    • H04N13/10—Processing, recording or transmission of stereoscopic or multi-view image signals
    • H04N13/106—Processing image signals
    • H04N13/111—Transformation of image signals corresponding to virtual viewpoints, e.g. spatial image interpolation
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
    • H04N13/20—Image signal generators
    • H04N13/204—Image signal generators using stereoscopic image cameras
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
    • H04N13/20—Image signal generators
    • H04N13/282—Image signal generators for generating image signals corresponding to three or more geometrical viewpoints, e.g. multi-view systems
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2200/00—Indexing scheme for image data processing or generation, in general
    • G06T2200/21—Indexing scheme for image data processing or generation, in general involving computational photography
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10052—Images from lightfield camera
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
    • H04N13/20—Image signal generators
    • H04N13/204—Image signal generators using stereoscopic image cameras
    • H04N13/243—Image signal generators using stereoscopic image cameras using three or more 2D image sensors
    • H—ELECTRICITY
    • H04—ELECTRIC COMMUNICATION TECHNIQUE
    • H04N—PICTORIAL COMMUNICATION, e.g. TELEVISION
    • H04N13/00—Stereoscopic video systems; Multi-view video systems; Details thereof
    • H04N2013/0074—Stereoscopic image analysis
    • H04N2013/0081—Depth or disparity estimation from stereoscopic image signals

Abstract

The present invention relates to the image processing equipment and image processing method for the anaglyph that multiple viewpoints can be generated with high-precision and small amount of calculation.Correlation generation unit uses the capture images captured from multiple viewpoints to calculate the relevance values from the capture images to capture of multiple viewpoints.Based on relevance values, generation unit generation includes the anaglyph of the anaglyph of the parallax value of the standard viewpoint as one of multiple viewpoints and the parallax value including occlusion area, the Hierarchical Disparity image as various criterion viewpoint.Present invention could apply to the image processing equipments such as light-field camera.

Description

Image processing equipment and image processing method
Technical field
This disclosure relates to image processing equipment and image processing method, relate more specifically to make it possible to small amount of calculation The image processing equipment and image processing method of the anaglyph of high accuracy generation different points of view.
Background technology
Recently the imaging device for being referred to as light-field camera etc. attracts attention.Such imaging device is obtained by imaging unit The capture images of viewpoint, the capture images (view interpolation) of virtual view are generated according to capture images, and use these captures The capture images of image generation virtual focus point etc. (for example, see patent document 1).
Such imaging device generates the image of virtual view as follows：By generating the capture images pair with viewpoint The anaglyph (depth map) answered, the capture of virtual view is generated using the anaglyph of respective viewpoints according to the capture images Image (image registration), and the capture images are combined.Therefore, in order to generate the high-precision capture figure of virtual view Picture, it is necessary that generation, which has high-precision anaglyph,.It should be noted that anaglyph is that have as the phase in capture images Answer the image of the pixel value of the parallax value of pixel.Here, the image positioned at the object compared with distal side has a less parallax value, and position There is larger parallax value in the image of the object compared with nearside.
Reference listing
Patent document
Patent document 1：JP 4706882 B2
The content of the invention
The problem to be solved in the present invention
The method for generating the anaglyph of viewpoint can be following method：According to the ginseng as the single viewpoint as reference The capture images generation anaglyph of viewpoint is examined, and the viewpoint in addition to reference view is estimated according to the anaglyph generated Anaglyph.
However, with it, when for example being blocked in the capture images of reference view, it is impossible to accurately estimation ginseng The parallax value of the occlusion area of viewpoint is examined, so as to which can not generate viewpoint in addition to reference view has high-precision disparity map Picture.It should be noted that in this manual, the occlusion area of reference view represents to watch from reference view but from except ginseng Examine the region that the viewpoint outside viewpoint can watch.
Alternatively, the method for generating the anaglyph of viewpoint can be following method：According to the capture figure of each viewpoint As generation anaglyph.With it, high-precision anaglyph can be generated, but need huge amount of calculation.
In view of said circumstances is made, the disclosure makes it possible to generate different points of view in high precision with small amount of calculation the disclosure Anaglyph.
The solution of problem
Image processing equipment according to the first aspect of the disclosure is following image processing equipment, and it includes：Computing unit, The computing unit calculates the relevance values of the capture images of the multipair viewpoint using the capture images of multipair viewpoint；It is and raw Into unit, the generation unit generates the different levels of reference view based on the relevance values calculated by the computing unit Anaglyph, the anaglyph is the viewpoint anaglyph that is formed by the parallax value of the reference view and by blocked area What the parallax value in domain was formed blocks anaglyph, and the reference view is one of described multipair viewpoint.
Set according to the image processing method of the first aspect of the disclosure and the image procossing of the first aspect according to the disclosure It is standby compatible.
In the first aspect of the disclosure, the capture figure of the multipair viewpoint is calculated using the capture images of multipair viewpoint The relevance values of picture, and based on relevance values come generate the viewpoint anaglyph formed by the parallax value of the reference view with And anaglyph is blocked by what the parallax value of occlusion area was formed, it is described as the anaglyph of the different levels of reference view Reference view is one of described multipair viewpoint.
Image processing equipment according to the second aspect of the disclosure is following image processing equipment, and it includes：Computing unit, The computing unit calculates the relevance values of the capture images of the multipair viewpoint using the capture images of multipair viewpoint；And Generation unit, pattern of the generation unit based on the relevance values for making mutually to tackle make a reservation for related to the multipair viewpoint The table of connection generates viewpoint anaglyph, the viewpoint anaglyph according to as with it is described it is predetermined to highest relevance values pair The highest correlation parallax value for the parallax value answered and generate, the viewpoint anaglyph is formed by the parallax value of reference view, institute Reference view is stated as one of described multipair viewpoint.
Set according to the image processing method of the second aspect of the disclosure and the image procossing of the second aspect according to the disclosure It is standby compatible.
In the second aspect of the disclosure, the capture figure of the multipair viewpoint is calculated using the capture images of multipair viewpoint The relevance values of picture；Pattern based on the relevance values for making mutually to tackle with the multipair viewpoint it is predetermined to associated table come Generate viewpoint anaglyph, the viewpoint anaglyph according to as with it is described it is predetermined to the corresponding parallax of highest relevance values The highest correlation parallax value of value and generate, the viewpoint anaglyph is formed by the parallax value of reference view, described with reference to regarding Point is one of described multipair viewpoint.
Can also be by making computer perform journey according to the image processing equipment of the first aspect of the disclosure and second aspect Sequence is realized.
In addition, in order to realize the image processing equipment of first aspect and second aspect according to the disclosure, treat by computer The program of execution by some transmission medium and can provide, or can provide the program of record on the recording medium.
The effect of the present invention
, can be with the high-precision anaglyph of small amount of calculation generation different points of view according to the disclosure.
Brief description of the drawings
Fig. 1 is the exemplary construction of the embodiment for the imaging device for showing the image processing equipment as the application disclosure Block diagram.
Fig. 2 is the perspective view of the exemplary construction of imaging array.
Fig. 3 is the block diagram for the exemplary construction for showing detection unit.
Fig. 4 is the figure for showing 16 viewpoints pair corresponding with correlation generation unit.
Fig. 5 is the figure for explaining the correlation detection for treating to be performed by correlation generation unit.
Fig. 6 is the figure of the example for the part for showing current pixel region.
Fig. 7 is the figure for the example for showing relevance values information.
Fig. 8 is the figure for the multigroup highest correlation parallax value for showing current pixel region.
Fig. 9 is the figure for other group of highest correlation parallax value for showing current pixel region.
Figure 10 is the figure for the example for showing layered sheet.
Figure 11 is the figure of the example of the anaglyph of the first level and the second level that show reference view.
Figure 12 is the figure of the example for the anaglyph for showing the viewpoint in addition to reference view.
Figure 13 is the figure of the example for the conventional parallax image for showing the viewpoint in addition to reference view.
Figure 14 is the figure for the example for showing virtual view.
Figure 15 is the figure for explaining the capture images of generation virtual view.
Figure 16 is the figure for explaining the capture images of generation virtual view.
Figure 17 is the figure for explaining the capture images of generation virtual view.
Figure 18 is the figure for explaining the capture images of generation virtual view.
Figure 19 is the figure for explaining the capture images of generation virtual view.
Figure 20 is the figure for explaining the capture images of generation virtual view.
Figure 21 is the flow chart for explaining imaging.
Figure 22 is the perspective view of another exemplary construction of imaging array.
Figure 23 is the block diagram of the example arrangement for the hardware for showing computer.
Embodiment
<Embodiment>
(exemplary construction of the embodiment of imaging device)
Fig. 1 is the exemplary construction of the embodiment for the imaging device for showing the image processing equipment as the application disclosure Block diagram.
Imaging device 10 shown in Fig. 1 includes imaging array 12 and graphics processing unit 13.Imaging device 10 obtains difference The capture images of viewpoint, and the capture images of virtual focus point are generated as refocusing image.
Specifically, the imaging array 12 of imaging device 10 is formed by the imaging unit of two-dimensional arrangement, each imaging unit bag Imaging len, image device such as CCD (charge coupling device) or CMOS (complementary metal oxide semiconductor) and A/D is included to turn Parallel operation.
Each imaging unit in imaging array 12 is received incident by the imaging len from the object with image device Light, and A/D conversions are carried out to the analog signal obtained due to light-receiving by using A/D converter and obtain capture images. By this way, imaging array 12 obtains the capture images of different points of view using each imaging unit.Imaging array 12 is to image Processing unit 13 provides the capture images of each viewpoint obtained.
Graphics processing unit 13 is formed by such as LSI (large scale integrated circuit).It is single that graphics processing unit 13 includes detection Member 21, virtual visual point image generation unit 22 and refocusing image generation unit 23.
Detection unit 21 be based not only on from imaging array 12 provide reference view capture images with except reference view it Relevance values between the capture images of outer viewpoint, and between the capture images based on the viewpoint in addition to reference view Relevance values detect parallax value.As a result, detect regarding for the parallax value of reference view and the occlusion area of reference view Difference.Explained to simplify, only exist an occlusion area of reference view in the present embodiment.
Detection unit 21 generates the anaglyph of two levels of reference view as follows：By will be by reference view The anaglyph that is formed of parallax value be arranged to the anaglyph (viewpoint anaglyph) of the first level, and will be by reference view The anaglyph that is formed of parallax value of occlusion area be arranged to the anaglyph (blocking anaglyph) of the second level.Detection is single Then member 21 provides the anaglyph of two levels of reference view to virtual visual point image generation unit 22.
Virtual visual point image generation unit 22 is to the disparity map of two levels of reference view provided from detection unit 21 As being stored.The anaglyph of two levels of stored reference view is used, virtual visual point image generation unit 22 is given birth to Into the anaglyph of the viewpoint in addition to reference view.
The anaglyph of each viewpoint stored and the capture figure of each viewpoint provided from imaging array 12 are provided Picture, virtual visual point image generation unit 22 generate the capture images of the virtual view in addition to viewpoint corresponding with imaging unit. Virtual visual point image generation unit 22 is provided as super multiple views (ultra-multiview) to refocusing image generation unit 23 The capture images of the capture images of the virtual view of image and each viewpoint provided from imaging array 12.
Refocusing image generation unit 23 is used as virtual focus point image generation unit, and use is given birth to from virtual visual point image The capture images of virtual focus point are generated as refocusing image into the super multi-view image of the offer of unit 22.Refocusing image is given birth to Generated refocusing image is exported into unit 23.
(exemplary construction of imaging array)
Fig. 2 be show the imaging array 12 shown in Fig. 1 exemplary construction scheme and be from subject side watch when The perspective view of imaging array 12.
In the imaging array 12 shown in Fig. 2,9 imaging unit 41-0 to 41-8 are disposed in low rate of expanding with heat and contract with cold Substrate planar surface 40 on, wherein three imaging units in 9 imaging unit 41-0 to 41-8 are by be equidistantly spaced from In every row and each column.As shown in Fig. 2 each imaging unit in imaging unit 41-0 to imaging unit 41-8 includes imaging len 42.In the present embodiment, viewpoint corresponding with imaging unit 41-0 is arranged to reference view.
In the following description, the vertical straight line that will extend through imaging unit 41-0 corresponding with reference view center It is referred to as angle θ with the angle being connected between imaging unit 41-0 center and the straight line at imaging unit 41-1 to 41-8 each center 1 to θ 8.Angle θ 1 to θ 8 respectively 45 degree, 360 degree, 315 degree, 90 degree, 270 degree, 135 degree, 180 degree and 225 degree.
The center for connecting imaging unit 41-0 and the length of the straight line at imaging unit 41-1 to 41-8 each center are claimed For distance L1 to L8.
(exemplary construction of detection unit)
Fig. 3 is the block diagram for the exemplary construction for showing the detection unit 21 shown in Fig. 1.
As shown in figure 3, detection unit 21 includes correlation generation unit 61-1 to 61-16, high correlation detection unit 62- 1 to 62-16, generation unit 63 and database 64.
Correlation generation unit 61-1 obtain from imaging unit 41-0 provide reference view capture images and from into The capture images provided as unit 41-2.Correlation generation unit 61-1 selects a left side for the capture images positioned at reference view first Then current pixel region is shifted one by the pixel region at upper angle every time as current pixel region along raster scanning direction Pixel.Pixel region can be formed as desired shape.
Correlation generation unit 61-1 is used as computing unit, and detection current pixel region is with providing from imaging unit 41-2 Correlation between reference pixel region corresponding to current pixel region in capture images, and obtain as correlation detection Result relevance values.For example, detection correlation method can according to phase difference detection relevance values method for example Block matching method, or it is related to detect according to the contrast obtained due to combination current pixel region and reference pixel region The method of property value.
Correlation generation unit 61-1 is also based on the distance between current pixel region and reference pixel region generation parallax Value.Correlation generation unit 61-1 and then provided to high correlation detection unit 62-1 represent the parallax value in current pixel region with The relevance values information of relation between relevance values corresponding to the parallax value.
Correlation generation unit 61-2 obtain from imaging unit 41-0 provide reference view capture images and from into The capture images provided as unit 41-1.Correlation generation unit 61-2 is held in a manner of with correlation generation unit 61-1 identicals Row correlation detection, and provide relevance values information to high correlation detection unit 62-2.
Correlation generation unit 61-3 obtain from imaging unit 41-0 provide reference view capture images and from into The capture images provided as unit 41-4.Correlation generation unit 61-3 is held in a manner of with correlation generation unit 61-1 identicals Row correlation detection, and provide relevance values information to high correlation detection unit 62-3.
Correlation generation unit 61-4 obtain from imaging unit 41-0 provide reference view capture images and from into The capture images provided as unit 41-6.Correlation generation unit 61-4 is held in a manner of with correlation generation unit 61-1 identicals Row correlation detection, and provide relevance values information to high correlation detection unit 62-4.
Correlation generation unit 61-5 obtain from imaging unit 41-0 provide reference view capture images and from into The capture images provided as unit 41-7.Correlation generation unit 61-5 is held in a manner of with correlation generation unit 61-1 identicals Row correlation detection, and provide relevance values information to high correlation detection unit 62-5.
Correlation generation unit 61-6 obtain from imaging unit 41-0 provide reference view capture images and from into The capture images provided as unit 41-8.Correlation generation unit 61-6 is held in a manner of with correlation generation unit 61-1 identicals Row correlation detection, and provide relevance values information to high correlation detection unit 62-6.
Correlation generation unit 61-7 obtain from imaging unit 41-0 provide reference view capture images and from into The capture images provided as unit 41-5.Correlation generation unit 61-7 is held in a manner of with correlation generation unit 61-1 identicals Row correlation detection, and provide relevance values information to high correlation detection unit 62-7.
Correlation generation unit 61-8 obtain from imaging unit 41-0 provide reference view capture images and from into The capture images provided as unit 41-3.Correlation generation unit 61-8 is held in a manner of with correlation generation unit 61-1 identicals Row correlation detection, and provide relevance values information to high correlation detection unit 62-8.
In the above described manner, correlation generation unit 61-1 to 61-8 generations are with the capture images of reference view and except reference The related relevance values information of the capture images of viewpoint outside viewpoint.Therefore, it is possible to be joined according to the relevance values infomation detection Examine the parallax value of viewpoint.
Correlation generation unit 61-9 obtain from imaging unit 41-2 provide capture images and from imaging unit 41- The capture images that 2 adjacent imaging unit 41-1 are provided.Correlation generation unit 61-9 then with correlation generation unit 61- 1 identical mode determines current pixel region.
Correlation generation unit 61-9 also detects the current pixel region in the capture images provided from imaging unit 41-2 Corresponding pixel region reference pixel area corresponding with the current pixel region in the capture images provided from imaging unit 41-1 Correlation between domain, and obtain the relevance values of the result as correlation detection.Correlation generation unit 61-9 also bases Parallax value is generated in the distance between pixel region corresponding with current pixel region and reference pixel region.Correlation generates Then unit 61-9 provides relevance values information to high correlation detection unit 62-9.
Correlation generation unit 61-10 obtain from imaging unit 41-1 provide capture images and from imaging unit The capture images that imaging unit 41-4 adjacent 41-1 is provided.Correlation generation unit 61-10 with correlation generation unit 61- 9 identical modes perform correlation detection, and provide relevance values information to high correlation detection unit 62-10.
Correlation generation unit 61-11 obtain from imaging unit 41-4 provide capture images and from imaging unit The capture images that imaging unit 41-6 adjacent 41-4 is provided.Correlation generation unit 61-11 with correlation generation unit 61- 9 identical modes perform correlation detection, and provide relevance values information to high correlation detection unit 62-11.
Correlation generation unit 61-12 obtain from imaging unit 41-6 provide capture images and from imaging unit The capture images that imaging unit 41-7 adjacent 41-6 is provided.Correlation generation unit 61-12 with correlation generation unit 61- 9 identical modes perform correlation detection, and provide relevance values information to high correlation detection unit 62-12.
Correlation generation unit 61-13 obtain from imaging unit 41-7 provide capture images and from imaging unit The capture images that imaging unit 41-8 adjacent 41-7 is provided.Correlation generation unit 61-13 with correlation generation unit 61- 9 identical modes perform correlation detection, and provide relevance values information to high correlation detection unit 62-13.
Correlation generation unit 61-14 obtain from imaging unit 41-8 provide capture images and from imaging unit The capture images that imaging unit 41-5 adjacent 41-8 is provided.Correlation generation unit 61-14 with correlation generation unit 61- 9 identical modes perform correlation detection, and provide relevance values information to high correlation detection unit 62-14.
Correlation generation unit 61-15 obtain from imaging unit 41-5 provide capture images and from imaging unit The capture images that imaging unit 41-3 adjacent 41-5 is provided.Correlation generation unit 61-15 with correlation generation unit 61- 9 identical modes perform correlation detection, and provide relevance values information to high correlation detection unit 62-15.
Correlation generation unit 61-16 obtain from imaging unit 41-3 provide capture images and from imaging unit The capture images that imaging unit 41-2 adjacent 41-3 is provided.Correlation generation unit 61-16 with correlation generation unit 61- 9 identical modes perform correlation detection, and provide relevance values information to high correlation detection unit 62-16.
In the above described manner, correlation generation unit 61-9 to 61-16 generations and the capture of the viewpoint in addition to reference view The related relevance values information of image.Therefore, it is possible to regarding for the occlusion area according to the relevance values infomation detection reference view Difference.
High correlation detection unit 62-1 is based on related from the current pixel region that correlation generation unit 61-1 is provided Relevance values information detects the highest relevance values in current pixel region.If the highest correlation in current pixel region Value is equal to or more than threshold value, then high correlation detection unit 62-1 provides regard corresponding with highest relevance values to generation unit 63 The parallax value (is referred to as highest correlation parallax value) by difference below.
On the other hand, if highest relevance values are less than threshold value, high correlation detection unit 62-1 is to generation unit 63 The invalidity information for representing that the parallax value in current pixel region is invalid is provided.It should be noted that threshold value as used herein is for example right Same image performs minimum relatedness value during correlation detection.
Similar to high correlation detection unit 62-1, high correlation detection unit 62-2 to 62-16 is based on from each correlation Property the relevance values information that provides of generation unit 61-2 to 61-16 provide highest correlation parallax value or invalid to generation unit 63 Property information.
Generation unit 63 from database 64 read layered sheet, in the layered sheet with and correlation generation unit 61-1 extremely Parallax value pattern corresponding to the highest relevance values of 16 viewpoints pair corresponding to 61-16 with and each level parallax value it is corresponding To associated.Generation unit 63 is based on the table and the current pixel area provided from high correlation detection unit 62-1 to 62-16 The related highest correlation parallax value in domain or invalidity information identify pair at each level.
Generation unit 63 calculate at each level identified to highest correlation parallax value average value simultaneously And store parallax value of the average value as current pixel region.Generation unit 63 and then by by all pictures at each level The parallax value in plain region is combined and generates anaglyph, and provides disparity map to virtual visual point image generation unit 22 Picture.
The storage of hierarchically table of database 64.
In the following description, when being not necessarily to special mutually differentiation correlation generation unit 61-1 to 61-16, by correlation Property generation unit 61-1 to 61-16 is referred to as correlation generation unit 61.Similarly, by high correlation detection unit 62-1 extremely 62-16 is referred to as high correlation detection unit 62.
(explanation of viewpoint pair corresponding with correlation generation unit)
Fig. 4 is to show 16 viewpoints pair corresponding with the correlation generation unit 61-1 to 61-16 shown in Fig. 3 Figure.
In the example depicted in fig. 4, imaging device 10 capture thereon with alphabetical A to Z background image, Yi Ji A football before background be present.Same case is applied to the Fig. 5 to Fig. 9 described afterwards.
As shown in figure 4, correlation generation unit 61-1 with by reference view corresponding with imaging unit 41-0 and and Viewpoint corresponding to imaging unit 41-2 formed to corresponding (below by this to being referred to as to #1).Correlation generation unit 61-2 is with leading to Cross reference view and with imaging unit 41-1 corresponding to viewpoint formed to corresponding (below by this to being referred to as to #2).Correlation Generation unit 61-3 with by reference to viewpoint and with imaging unit 41-4 corresponding to viewpoint formed to corresponding (below that this is right Referred to as to #3).
Correlation generation unit 61-4 and pair by reference to viewpoint and viewpoint corresponding with imaging unit 41-6 formation Corresponding (below by this to being referred to as to #4).Correlation generation unit 61-5 with by reference to viewpoint and with 41-7 pairs of imaging unit The viewpoint answered formed to corresponding (below by this to being referred to as to #5).Correlation generation unit 61-6 with by reference to viewpoint and With imaging unit 41-8 corresponding to viewpoint formed to corresponding (below by this to being referred to as to #6).
Correlation generation unit 61-7 and pair by reference to viewpoint and viewpoint corresponding with imaging unit 41-5 formation Corresponding (below by this to being referred to as to #7).Correlation generation unit 61-8 with by reference to viewpoint and with 41-3 pairs of imaging unit The viewpoint answered formed to corresponding (below by this to being referred to as to #8).
Correlation generation unit 61-9 with by viewpoint corresponding with imaging unit 41-2 and with 41-1 pairs of imaging unit The viewpoint answered formed to corresponding (below by this to being referred to as to #9).Correlation generation unit 61-10 with by and imaging unit Viewpoint corresponding to 41-1 and with imaging unit 41-4 corresponding to viewpoint formed to corresponding (below by this to being referred to as to #10). Correlation generation unit 61-11 with by viewpoint corresponding with imaging unit 41-4 and with imaging unit 41-6 corresponding to viewpoint Formed to corresponding (below by this to being referred to as to #11).
Correlation generation unit 61-12 with by viewpoint corresponding with imaging unit 41-6 and with 41-7 pairs of imaging unit The viewpoint answered formed to corresponding (below by this to being referred to as to #12).Correlation generation unit 61-13 with by and imaging unit Viewpoint corresponding to 41-7 and with imaging unit 41-8 corresponding to viewpoint formed to corresponding (below by this to being referred to as to #13). Correlation generation unit 61-14 with by viewpoint corresponding with imaging unit 41-8 and with imaging unit 41-5 corresponding to viewpoint Formed to corresponding (below by this to being referred to as to #14).
Correlation generation unit 61-15 with by viewpoint corresponding with imaging unit 41-5 and with 41-3 pairs of imaging unit The viewpoint answered formed to corresponding (below by this to being referred to as to #15).Correlation generation unit 61-16 with by and imaging unit Viewpoint corresponding to 41-3 and with imaging unit 41-2 corresponding to viewpoint formed to corresponding (below by this to being referred to as to #16).
(explanation of correlation detection)
Fig. 5 is the figure for explaining the correlation detection for treating to be performed by correlation generation unit 61.
As shown in figure 5, such as its center the capture images of reference view coordinate (x0,y0) place small pixel area 80 when being current pixel region, and correlation generation unit 61-1 is based on coordinate (x0,y0) determine to catch from what imaging unit 41-2 was provided Obtain the coordinate (x at the center in the reference pixel region 81 in image1,y1)。
Specifically, with coordinate (x0,y0) initial value is used as, correlation generation unit 61-1 determines coordinate (x successively1,y1) be Along the direction of angle corresponding with imaging unit 41-2 (θ 2+180) degree or the displacement of polar curve direction and short distance Δ d2Equal amount Coordinate.Correlation between correlation generation unit 61-1 and then detection pixel region 80 and reference pixel region 81.
Correlation generation unit 61-2 to 61-8 performs correlation inspection in a manner of with correlation generation unit 61-1 identicals Survey.That is, correlation generation unit 61-i (i=2,3,4,5,6,7,8) determines reference pixel region 8i as follows Center coordinate (xi,yi)：With coordinate (x0,y0) initial value is used as, by coordinate (xi,yi) along angle (θ j+180) (j=1,4, 6,7,8,5,3) the direction displacement of degree is equal to distance, delta djIt is a small amount of.Correlation generation unit 61-i and then detection pixel region 80 With the correlation between the 8i of reference pixel region.
In the above described manner, with reference view to each phase in corresponding correlation generation unit 61-1 to 61-8 Closing property generation unit sets the reference pixel region in current pixel region, and detects current pixel region and reference pixel region Between correlation.
Meanwhile correlation generation unit 61-9 determines pixel region corresponding with current pixel region.Specifically, current The coordinate at the center of pixel region is coordinate (x0,y0) when, correlation generation unit 61-9 is determined successively：Initial value is coordinate (x0,y0), along direction displacement and the Δ d of angle corresponding with imaging unit 41-2 (θ 2+180) degree2The coordinate of equal amount be with The coordinate at the center of pixel region corresponding to current pixel region.
With coordinate (x0,y0) initial value is used as, correlation generation unit 61-9 is also determined successively：Edge and imaging unit 41-1 The direction displacement of corresponding angle (θ 1+180) degree and Δ d1The coordinate of equal amount is the coordinate at the center in reference pixel region. Correlation generation unit 61-9 then detect and current pixel region corresponding to phase between pixel region and reference pixel region Guan Xing.Correlation generation unit 61-10 to 61-16 performs correlation inspection in a manner of with correlation generation unit 61-9 identicals Survey.
In the above described manner, with the viewpoint in addition to reference view to corresponding correlation generation unit 61-9 to 61-16 In each correlation generation unit the reference in corresponding with current pixel region pixel region and current pixel region is set Pixel region.It is then detected that the and correlation corresponding to current pixel region between pixel region and reference pixel region.
It should be noted that Δ d1To Δ dsIt is the value defined in formula (1) as shown below.
[mathematical formulae 1]
L1/Δd1=L2/ Δs d2=L3/ Δs d3=L4/ Δs d4=L5/ Δs d5=L6/ Δs d6=L7/ Δs d7=L8/ Δs d8
...(1)
(explanation of layered sheet)
Fig. 6 is that the part for showing current pixel region in the capture images of the reference view shown in Fig. 5 is shown The figure of example.
As shown in fig. 6, in the capture images of the reference view shown in Fig. 5, current pixel region can be to be included in the back of the body The pixel region 101 of " J " part write in scape, include the pixel region 102 of " K " part, and including being hidden by football The pixel region 103 of " L " etc..
Fig. 7 is to show coordinate of the coordinate at the center of pixel region 101 to 103 for the center in current pixel region In the case of on the capture images shown in Fig. 5 relevance values information figure.
In the figure 7, abscissa represents the parallax value of relevance values information, and ordinate represents relevance values.
As shown in Fig. 7 left column, the coordinate at the center of pixel region 101 is the coordinate at the center in current pixel region In the case of in the relevance values information to #1 to #16, relevance values highest corresponding with the parallax value of background, and described Relevance values are equal to or more than threshold value.
That is, in the capture images shown in Fig. 5, the background of pixel region 101 at all viewpoints be present Image.Correspondingly, all highest relevance values in #1 to #16 are equal to or more than threshold value, and highest relevance values for the back of the body The parallax value of scape.
In the case where all highest correlation parallax values in #1 to #16 is the parallax values of background, the back of the body can be used The parallax value of scape generates the parallax value of first level in current pixel region.
As shown in Fig. 7 middle column, the coordinate at the center of pixel region 102 is the feelings at the center in current pixel region Under condition on to #1, to #5 to #8 and in #13 to #16 relevance values information, to the parallax value of background corresponding to it is related Property value highest, and the relevance values are equal to or more than threshold value.
That is, in the capture images shown in Fig. 5, by imaging unit 41-0, imaging unit 41-2, imaging unit The back of the body of pixel region 102 in 41-3, imaging unit 41-5, imaging unit 41-7 and the image of imaging unit 41-8 captures be present The image of scape.Correspondingly, it is equal to or more than threshold value to #1, to #5 to #8 and to the highest relevance values in #13 to #16, and And highest correlation parallax value is the parallax value of background.
Meanwhile on to #2 to #4, to #9 and in #12 relevance values information, highest relevance values are less than threshold Value.On to #10 and in #11 relevance values information, relevance values highest corresponding with the parallax value of football, and The relevance values are equal to or more than threshold value.
That is, in the capture images shown in Fig. 5, the image of the background of pixel region 102 is by by imaging unit 41- 1st, the image of the football in the image of imaging unit 41-4 and imaging unit 41-6 captures is hidden.Therefore, with it is single by imaging First 41-1, imaging unit 41-4 and imaging unit 41-6 capture pixel region 102 corresponding to image be different from by being imaged Unit 41-0, imaging unit 41-2, imaging unit 41-3, imaging unit 41-5, imaging unit 41-7 and imaging unit 41-8 Image corresponding to the pixel region 102 of capture.
Correspondingly, highest relevance values compare #2 to #4, it is to #9 and small to the threshold value in #12, to #2 to #4, to #9 with It is and corresponding with imaging unit 41-1, imaging unit 41-4 or imaging unit 41-6 and single with imaging unit 41-0, imaging to #12 First 41-2, imaging unit 41-3, imaging unit 41-5, imaging unit 41-7 or imaging unit 41-8 are corresponding.
Meanwhile the pixel region 102 with being captured by imaging unit 41-1, imaging unit 41-4 and imaging unit 41-6 is right The image answered is identical, shows football.Correspondingly, with imaging unit 41-1, imaging unit 41-4 and imaging unit 41-6 In two imaging units corresponding to #10 and in #11, highest relevance values are equal to or more than threshold value, and highest correlation Property parallax value be football parallax value.
As described above, the center of pixel region 102 coordinate for current pixel region center coordinate when, on The rectangle 121 represented by the dotted line in Fig. 8 surround to #1, to #5 to #8 and in #13 to #16 relevance values information, Highest correlation parallax value is the parallax value of background.In the correlation to #10 and to #11 on being surrounded by dashed rectangle 122 In value information, highest correlation parallax value is the parallax value of football.
In this case, it is all to corresponding highest correlation parallax value etc. with the viewpoint in addition to reference view In or more than with reference view to corresponding highest correlation parallax value, therefore be not the occlusion area of reference view Parallax value.Correspondingly, the parallax value using background and the highest correlation parallax value to #13 to #16 can generate current picture The parallax value of only first level in plain region, the parallax value of background be have reference view to highest correlation parallax value, And there is the parallax value of the background as highest correlation parallax value to #13 to #16.
As shown in Fig. 7 right row, the coordinate at the center of pixel region 103 is the coordinate at the center in current pixel region When, on to #1 to #13 and in #16 relevance values information, relevance values highest corresponding with the parallax value of football, And the relevance values are equal to or more than threshold value.
It is by imaging unit with 103 corresponding image of pixel region that is, in the capture images shown in Fig. 5 The image of football in the image that 41-0 to 41-2, imaging unit 41-4, imaging unit 41-6 and imaging unit 41-7 are captured. In addition, the image of football in the capture images of all viewpoints be present.Correspondingly, have in formation with imaging unit 41-0 extremely 41-2, imaging unit 41-4, imaging unit 41-6 and viewpoint corresponding to imaging unit 41-7 to #1 to #13 and To in #16, highest relevance values are equal to or more than threshold value and highest correlation parallax value is the parallax value of football.
Meanwhile on to #14 and in #15 relevance values information, relevance values corresponding with the parallax value of background Highest, and the relevance values are equal to or more than threshold value.
That is, in the capture images shown in Fig. 5, the image with 103 corresponding image of pixel region for background, By the image of background in each image of imaging unit 41-3, imaging unit 41-5 and imaging unit 41-8 capture be present.Phase Ying Di, with viewpoint corresponding with imaging unit 41-3, imaging unit 41-5 or imaging unit 41-8 to corresponding to # 14 and in #15, highest relevance values are equal to or more than threshold value and highest correlation parallax value is the parallax value of background.
As described above, in the case where the coordinate at the center of pixel region 103 is the coordinate at center in current pixel region, In the relevance values information to #1 to #13 and to 16 that the rectangle 141 on being represented by the dotted line in Fig. 9 surrounds, highest Correlation parallax value is the parallax value of football.In the relevance values letter to #14 and to #15 on being surrounded by dashed rectangle 142 In breath, highest correlation parallax value is the parallax value of background.
In this case, it is all to corresponding highest correlation parallax value ratio with the viewpoint in addition to reference view It is small to corresponding highest correlation parallax value with reference view.That is, occurs ginseng in current pixel region Examine blocking for viewpoint.
Correspondingly, using the viewpoint in addition to reference view to highest correlation parallax value in, than with ginseng Only second level in current pixel region can be generated by examining the parallax value small to corresponding highest correlation parallax value of viewpoint Parallax value.
In addition, it can be generated using the parallax value of background and to #9 to #13 and to #16 highest correlation parallax value The parallax value of only first level in current pixel region, the parallax value of background for reference view to highest correlation regard Difference, and to #9 to #13 and to parallax values of the #16 with the background as highest correlation parallax value.
As described above with described in Fig. 7 to Fig. 9, checking with according to parallax value mould corresponding with each pair of highest relevance values Formula determines whether to occur blocking for reference view in current pixel region.In this way it is possible to determine current pixel area The number of the level of the parallax value in domain and the parallax value to be used when generating the parallax value of each level.Correspondingly, dividing Layer table in, to each pair of highest relevance values corresponding to parallax value pattern with and each level parallax value it is corresponding to related Connection.
Figure 10 is the figure for the example for showing layered sheet.
As shown in Figure 10, in layered sheet, the minimum value and maximum of all highest correlation parallax values to #1 to #16 Fall pattern in preset range with as and the first level parallax value it is corresponding to #1 to #16 is associated.Correspondingly, exist The minimum value and maximum of all pairs of highest correlation parallax value fall the highest correlation parallax in preset range or all pairs In the case of being worth for a kind of parallax value, the parallax value of the first level is generated using all pairs of highest correlation parallax values.
To #1, to #5 to #8 and to the highest correlation parallax value in #13 to #16 be compared with the parallax value in distal side, to #10 Be compared with the relevance values of nearside and the highest relevance values of remaining centering to compare threshold value to the highest correlation parallax value in #11 Small pattern with to #1, it is to #5 to #8 and associated to #13 to #16, to #1, to #5 to #8 and to #13 to #16 with compared with The parallax value in distal side corresponds to, as corresponding with the first level right.
Similarly, it is compared with distal side to #1, to #2, to #6 to #9 and to the highest correlation parallax value in #14 to #16 Parallax value, to #11 to the highest correlation parallax value in #12 to be related compared with the parallax value of nearside and the highest of remaining centering Property the value pattern smaller than threshold value with and it is corresponding compared with the parallax value in distal side, as and the first level corresponding to to associated.
Similarly, to #1 to #3, to #7 to #10, to #15 and to the highest correlation parallax value in #16 be compared with distal side Parallax value, to #12 and to the highest correlation parallax value in #13 be compared with the parallax value of nearside and the highest phase of remaining centering The closing property value pattern smaller than threshold value with and compared with distal side parallax value it is corresponding, as and the first level corresponding to to being associated.
Similarly, to #1 to #4, to #8 to #11 and to the highest correlation parallax value in #16 be compared with distal side parallax Value, to #13 and to the highest correlation parallax value in #14 be compared with the parallax value of nearside and the highest relevance values of remaining centering The pattern smaller than threshold value with and compared with distal side parallax value it is corresponding, as and the first level corresponding to associated.
Similarly, to #1 to #5 and to the highest correlation parallax value in #9 to #12 be compared with the parallax value in distal side, to # 14 and be compared with the parallax value of nearside and the highest relevance values of remaining centering to compare threshold value to the highest correlation parallax value in #15 Small pattern with and compared with distal side parallax value it is corresponding, as and the first level corresponding to to associated.
Similarly, to #2 to #6 and to the highest correlation parallax value in #10 to #13 be compared with the parallax value in distal side, to # 15 and be compared with the parallax value of nearside and the highest relevance values of remaining centering to compare threshold value to the highest correlation parallax value in #16 Small pattern with and compared with distal side parallax value it is corresponding, as and the first level corresponding to to associated.
Similarly, to #3 to #7 and to the highest correlation parallax value in #11 to #14 be compared with the parallax value in distal side, to # 9 and be compared with the parallax value of nearside and the highest relevance values of remaining centering to compare threshold value to the highest correlation parallax value in #16 Small pattern with and compared with distal side parallax value it is corresponding, as and the first level corresponding to to associated.
Similarly, to #4 to #8 and to the highest correlation parallax value in #12 to #15 be compared with the parallax value in distal side, to # 9 and be compared with the parallax value of nearside and the highest relevance values of remaining centering to compare threshold value to the highest correlation parallax value in #10 Small pattern with and compared with distal side parallax value it is corresponding, as and the first level corresponding to to associated.
With reference view to highest correlation parallax value be compared with the parallax value in distal side in the case of, with above-mentioned side Formula use with highest correlation parallax value to highest correlation parallax value generate first as the parallax value compared with distal side The parallax value of level.
Meanwhile to #1 to #13 and to the highest correlation parallax value in #16 for compared with the parallax value of nearside and to #14 For the pattern of the parallax value compared with distal side and include with to the highest correlation parallax value in #15 to #1 to #8, to #10 and to # 10 of 11 to associated, to #1 to #8, to #10 and to #11 be with it is corresponding compared with the parallax value of nearside in first It is right corresponding to level.The pattern also with compared with the parallax value in distal side it is corresponding, as and the second level corresponding to to #14 and #15 is associated.
Similarly, it is compared with the parallax value of nearside and to #15 and to #16 to the highest correlation parallax value in #1 to #14 In highest correlation parallax value be parallax value compared with distal side pattern with including 10 to #1 to #8, to #11 and to #12 To associated, to #8, to #11 and to #12 be to #1 with it is corresponding compared with the parallax value of nearside in the first level pair Pair answered.The pattern also to compared with the parallax value in distal side it is corresponding, as and the second level corresponding to it is to #15 and related to #16 Connection.
Similarly, to #1 to #8 and to the highest correlation parallax value in #10 to #15 be compared with the parallax value of nearside, to # 9 and to the highest correlation parallax value in #16 be compared with distal side parallax value pattern with including to #1 to #8, to #12 and to # 10 of 13 to associated, to #1 to #8, to #12 and to #13 be with it is corresponding compared with the parallax value of nearside in the It is right corresponding to one level.The pattern also with compared with the parallax value in distal side it is corresponding, as and the second level corresponding to to #9 and #16 is associated.
Similarly, to #1 to #8 and to the highest correlation parallax value in #11 to #16 be compared with nearside parallax value and To #9 and to pattern that the highest correlation parallax value in #10 is the parallax value compared with distal side with including to #1 to #8, to #13 and To 10 of #14 to associated, to #8, to #13 and to #14 be to #1 with it is corresponding compared with the parallax value of nearside it is in It is right corresponding to first level.The pattern also with compared with the parallax value in distal side it is corresponding, as and the second level corresponding to to #9 It is associated with to #10.
Similarly, to #1 to #9 and to the highest correlation parallax value in #12 to #16 be compared with nearside parallax value and To #10 and to the highest correlation parallax value in #11 be compared with distal side parallax value pattern with including to #1 to #8, to #14 with And to 10 of #15 to associated, to #8, to #14 and to #15 be to #1 with corresponding compared with the parallax value of nearside in It is corresponding with the first level right.The pattern also with compared with the parallax value in distal side it is corresponding, as and the second level corresponding to to # 10 and #11 is associated.
Similarly, to #1 to #10 and to the highest correlation parallax value in #13 to #16 be compared with nearside parallax value with And for the pattern of the parallax value compared with distal side and include to #11 and to the highest correlation parallax value in #12 to #1 to #8, to #15 And to 10 of #16 to associated, to #8, to #15 and to #16 be to #1 with corresponding compared with the parallax value of nearside in It is corresponding with the first level right.The pattern also with compared with the parallax value in distal side it is corresponding, as and the second level corresponding to It is associated to #11 and to #12.
Similarly, to #1 to #11 and to the highest correlation parallax value in #14 to #16 be compared with nearside parallax value with And for the pattern of the parallax value compared with distal side and include to #12 and to the highest correlation parallax value in #13 to #1 to #9 and to # 10 of 16 to associated, to #1 to #9 and to #16 be with it is corresponding compared with the parallax value of nearside in the first level It is corresponding right.The pattern also with compared with the parallax value in distal side it is corresponding, as and the second level corresponding to to #12 and to #13 It is associated.
Similarly, to #1 to #12, to #15 and to the highest correlation parallax value in #16 be compared with nearside parallax value with And to #13 and to the highest correlation parallax value in #14 for the pattern of the parallax value compared with distal side and including 10 to #1 to #10 To associated, be to #1 to #10 with it is corresponding compared with the parallax value of nearside in corresponding with the first level right.The pattern is also With with compared with the parallax value in distal side it is corresponding, as and the second level corresponding to to #13 and #14 is associated.
With reference view to highest correlation parallax value be compared with the parallax value of nearside in the case of, with above-mentioned side Formula use with highest correlation parallax value to highest correlation parallax value generate first as the parallax value compared with nearside The parallax value of level.In addition, use with highest correlation parallax value to highest correlation parallax value as compared with distal side Parallax value generates the parallax value of second layer level.
In the example depicted in fig. 10, with the highest correlation parallax value as the parallax value compared with nearside in only 10 pairs associated with the first level.However, with all right of the highest correlation parallax value as the parallax value compared with nearside Can be associated with the first level.
(example of the first level of reference view and the anaglyph of the second level)
Figure 11 is the first level of the capture images for showing the reference view shown in Fig. 4 and the disparity map of the second level The figure of the example of picture.
As shown in figure 11, for the capture images 161 of reference view, the parallax formed by the parallax value of reference view is generated Anaglyph of the image 162 as the first level of reference view.In anaglyph 162, with the football in capture images 161 Corresponding region 162A parallax value is big, and region 162B corresponding with background parallax value is small.
In addition, as shown in figure 11, for the capture images 161 of reference view, generate by the occlusion area of reference view Anaglyph of the anaglyph 163 that parallax value is formed as the second level of reference view.In anaglyph 163, background Parallax value for it is corresponding with the region inwardly away from the border preset distance between the football in capture images 161 and background annular Region 163A parallax value.The parallax value of region 163B in anaglyph 163 in addition to the 163A of region is expression ineffectivity Value.
(example of the anaglyph of the viewpoint in addition to reference view)
Figure 12 is the figure of the example for the anaglyph for showing the viewpoint in addition to reference view, uses regarding shown in Figure 11 Difference image 162 and anaglyph 163 generate the anaglyph.
Example in Figure 12 illustrate only the imaging on the left of imaging unit 41-0 corresponding with reference view object The anaglyph of unit 41-5 viewpoint, and the viewpoint of imaging unit 41-4 on the right side of imaging unit 41-0 object Anaglyph.However, be not other viewpoints of reference view anaglyph it is identical with these anaglyphs.Shown in Figure 12 In view, abscissa represents the position of each pixel of (X-direction) in the horizontal direction, and ordinate represents parallax value.It is identical Situation is applied to the Figure 13 described afterwards.
As represented by the centre view in Figure 12, as the corresponding region of occlusion area in the capture images with reference view Anaglyph 162 and the parallax value that represents of anaglyph 163 be used as the parallax value of background and football.That is, disparity map As 162 and anaglyph 163 have identical parallax value in region corresponding with occlusion area.
Correspondingly, virtual visual point image generation unit 22 can accurately generate the capture with imaging unit 41-5 viewpoint Anaglyph 182 corresponding to image 181 and the 191 corresponding anaglyph of capture images with imaging unit 41-4 viewpoint 192。
That is, the region 162A's that virtual visual point image generation unit 22 passes through right shift anaglyph 162 regards Difference generates anaglyph 182, but can not accurately generate the parallax value of occlusion area, because in anaglyph Corresponding parallax value is not present in 162.
However, virtual visual point image generation unit 22 can also use anaglyph 163, so that correspondingly can be by making Anaglyph 182 is accurately generated by the use of the parallax value of anaglyph 163 as the parallax value of occlusion area.
Similarly, the parallax for the region 162A that virtual visual point image generation unit 22 passes through shifted left anaglyph 162 Value generates anaglyph 192, but can accurately generate occlusion area by using the parallax value of anaglyph 163 Parallax value.
On the other hand, in the situation of generation only anaglyph 162, as shown in the centre view in Figure 13, screening is not generated Keep off the parallax value in region.Because the screening in anaglyph 211 corresponding with capture images 181 is not present in anaglyph 162 Region 211A parallax value is kept off, so occlusion area 211A parallax value can not be accurately generated.
Similarly, because the screening in anaglyph 212 corresponding with capture images 191 is not present in anaglyph 162 Region 212A parallax value is kept off, so occlusion area 212A parallax value can not be accurately generated.
(example of virtual view)
Figure 14 is the figure for the example for showing virtual view.
In the example depicted in fig. 14, virtual view be with positioned at imaging unit 41-2, imaging unit 41-3, imaging unit Viewpoint corresponding to the virtual image unit 231 of 41-7 and imaging unit 41-8 central spot.
(explanation of the method for the capture images of generation virtual view)
Figure 15 to Figure 20 is the figure for explaining the capture images of the virtual view shown in generation Figure 14.
As shown in figure 15, in the capture images 250 for generating the virtual view shown in Figure 14, virtual visual point image generation is single Member 22 uses the image 251 captured by imaging unit 41-3, the image 252 captured by imaging unit 41-2, by imaging unit 41- The image 253 of 8 captures and the image 254 captured by imaging unit 41-7.Virtual visual point image generation unit 22 also use with Anaglyph 261 to 264 corresponding to capture images 251 to 254.
Specifically, as shown in figure 16, virtual visual point image generation unit 22 is first by capture images 251 and anaglyph 261 generate the capture images 281 of virtual view.Specifically, virtual visual point image generation unit 22 is by by capture images 251 Each pixel shift amount equal with being worth as follows generate the capture images 281 of virtual view, the value is based on anaglyph 261 In pixel parallax value and position for example between reference view, imaging unit 41-2 viewpoint and virtual view close System.
As shown in figure 17, virtual visual point image generation unit 22 then by with generation the identical of capture images 281 in a manner of, make The capture images 282 of virtual view are generated with capture images 252 and anaglyph 262.As shown in figure 18, virtual visual point image Generation unit 22 is also generated in a manner of with the generation identical of capture images 281, using capture images 253 and anaglyph 263 The capture images 283 of virtual view.As shown in figure 19, virtual visual point image generation unit 22 is also with generating capture images 281 Identical mode, the capture images 284 for generating using capture images 254 and anaglyph 264 virtual view.
As shown in figure 20, virtual visual point image generation unit 22 and then the progress of capture images 281 to 284 by virtual view Combination, and the final capture images 250 of generation virtual view.
(treating the explanation of processing performed by imaging device)
Figure 21 is the flow chart for explaining the imaging for treating to be performed by the imaging device 10 shown in Fig. 1.
In Figure 21 step S11, the imaging array 12 of imaging device 10 obtains each of imaging unit 41-0 to 41-8 The capture images of viewpoint, and provide capture images to detection unit 21.In step s 12, the correlation life of detection unit 21 The pixel region of the capture images of reference view is provided as current pixel region into unit 61 (Fig. 3).
Specifically, in the capture images of reference view, correlation generation unit 61 by the pixel region of preliminary dimension from Upper left shifts the amount equivalent to a pixel along raster scanning direction every time, so as to set the pixel region of preliminary dimension. Top left pixel region is provided as the current pixel region during first by correlation generation unit 61 in step s 12, and And pixel region is set according to the order of raster scanning during afterwards in step s 12.
In step s 13, correlation generation unit 61 performs correlation to capture images relative to current pixel region Detection, and generation relevance values information.Then correlation generation unit 61 provides correlation to high correlation detection unit 62 Value information.
In step S14, high correlation detection unit 62 is based on the relevance values letter provided from correlation generation unit 61 Cease to detect the highest relevance values in current pixel region.In step S15, high correlation detection unit 62 determines to be detected Highest relevance values whether be equal to or more than threshold value.
If determining that highest relevance values are equal to or more than threshold value in step S15, high correlation is examined in step s 16 Survey unit 62 and export highest correlation parallax value to generation unit 63, the processing is then proceeded by step S18.
It is high related in step S17 if determining that highest relevance values are not equal to or more than threshold value in step S15 Property detection unit 62 export invalidity information to generation unit 63, the processing then proceeded by step S18.
In step S18, generation unit 63 reads layered sheet from database 64.In step S19, generation unit 63 is based on Layered sheet and the highest correlation parallax value provided from each high correlation detection unit 62 or invalidity information identify position In pair of each level.
In step S20, generation unit 63 calculated for each level identified in step S19 to highest it is related The average value of property parallax value.In the step s 21, generation unit 63 is by the highest phase of each level calculated in step S20 The average value of closing property parallax value stores the parallax as the current pixel region in the anaglyph of each level of reference view Value.
In step S22, correlation generation unit 61 determines whether all pixels region is arranged into current pixel area Domain.It is described to the processing returns to step if determining that all pixels region not yet is arranged into current pixel region in step S22 S12, repeat step S12 to S22 process by all pixels region untill current pixel region is arranged to.
On the other hand, if determining that all pixels region is defined as into current pixel region in step S22, Generation unit 63 exports the first level and the of stored reference view to virtual visual point image generation unit 22 in step S23 The anaglyph of two levels.
In step s 24, using the first level of the reference view provided from generation unit 63 and the parallax of the second level Image, virtual visual point image generation unit 22 generate the anaglyph of the viewpoint in addition to reference view.
In step s 25, virtual visual point image generation unit 22 is using the anaglyph of each viewpoint and from imaging array The capture images of the 12 each viewpoints provided generate the capture images of virtual view.Virtual visual point image generation unit 22 is to again Focusedimage generation unit 23 provides the capture images of the virtual view as super multi-view image and provided from imaging array 12 Each viewpoint capture images.
In step S26, refocusing image generation unit 23 is super using being provided from virtual visual point image generation unit 22 Multi-view image generates the capture images of virtual focus point as refocusing image.The output of refocusing image generation unit 23 is given birth to Into refocusing image.
In the above described manner, imaging device 10 is generated by the anaglyph that the parallax value of reference view is formed as the first level Anaglyph, and generation by the anaglyph that the parallax value of the occlusion area of reference view is formed regarding as the second level Difference image.Correspondingly, the anaglyph of the viewpoint in addition to reference view can be accurately generated.
Imaging device 10 generates the disparity map of the viewpoint in addition to reference view always according to the anaglyph of reference view Picture.Correspondingly, each viewpoint is being generated according to capture images for generating the amount of calculation ratio needed for the anaglyph of each viewpoint Anaglyph in the case of needed for amount of calculation it is small.In addition, for afterwards using in the processing of the anaglyph of each viewpoint Use need store only with reference to viewpoint anaglyph.Accordingly, it is desirable to the number of the anaglyph of storage is small.The opposing party Face, in the case where generating the anaglyph of each viewpoint according to capture images, it is necessary to store the anaglyph of all viewpoints.
Further, imaging device 10 is in a hierarchical manner to the anaglyph that is formed by the parallax value of reference view and by joining The anaglyph for examining the parallax value formation of the occlusion area of viewpoint is stored.Correspondingly, reference view can efficiently be stored Parallax value and reference view occlusion area parallax value.It is easily and accurate that imaging device 10 can also be based on layered sheet Ground selects the highest correlation parallax value to be used when generating anaglyph.
(another exemplary construction of imaging array)
Figure 22 is the figure for another exemplary construction for showing the imaging array 12 shown in Fig. 1 and watched from subject side When imaging array 12 perspective view.
In the imaging array 12 shown in Figure 22,7 imaging unit 301-0 to 301-6 are arranged on planar surface 40 So that imaging unit 301-1 to 301-6 is located at using imaging unit 301-0 as each angle of the regular hexagon at center.
In the example shown in Figure 22, imaging unit 301-1 to 301-6 is located at each angle of regular hexagon.However, imaging Unit 301-1 to 301-6 can be located at each angle of hexagon.That is, the phase in imaging unit 301-1 to 301-6 The distance between adjacent imaging unit can be with unequal.
In addition, the number and location for forming the imaging unit of imaging array 12 are not restricted to showing shown in Fig. 2 and Figure 22 Number and location in example.
Explain the only one occlusion area of reference view in the present embodiment be present to simplify.However, it is possible in the presence of not An only occlusion area.That is, capture images can be to show the image of the object before background.In such situation Under, the number of the level of the anaglyph of reference view is by increasing the number that " 1 " obtained to the number of occlusion area.
(using the explanation of the computer of the disclosure)
A series of processing described above can be performed by above-mentioned hardware, it is also possible to be performed by software.Logical When crossing a series of software execution processing, the program for forming software is installed into computer.Here, the computer can be with It is general for including the computer of specialized hardware or various functions being able to carry out when can be and be provided with various programs wherein People's computer.
Figure 23 is the block diagram of the example arrangement for the hardware for showing the computer that a series of above-mentioned processing are performed according to program.
In a computer, (arbitrary access is deposited by CPU (CPU) 601, ROM (read-only storage) 602 and RAM Reservoir) 603 it is connected with each other by bus 604.
Input/output interface is further attached to bus 604.Imaging unit 606, input block 607, output unit 608, Memory cell 609, communication unit 610 and driver 611 are connected with input/output interface 605.
Imaging unit 606 is formed by the grade of imaging array 12 and obtains the capture images of different points of view.Input block 607 by The formation such as keyboard, mouse, microphone.Output unit 608 is formed by display, loudspeaker etc..Memory cell 609 is by hard disk, non- The formation such as volatile memory.Communication unit 610 is formed by network interface etc..Driver 611 drives removable media 612 for example Disk, CD, magneto-optic disk or semiconductor memory.
In the computer with said structure, CPU 601 will deposit for example, by input/output interface 605 and bus 604 Store up the program in memory cell 609 and be loaded into RAM 603 simultaneously configuration processor so that perform a series of above-mentioned processing.
The program for treating to perform (CPU 601) execution by computer can be recorded in for example as encapsulation medium to be supplied Removable media 612 on.Alternatively, program can by wired or wireless transmission medium such as LAN, internet or Digital satellite broadcasting provides.
In a computer, input/output interface can be passed through when removable media 612 is arranged on driver 611 605 are attached to program in memory cell 609.Journey can also be received by wired or wireless transmission medium by communication unit 610 Sequence and program is attached in memory cell 609.Alternatively, program can be attached to ROM 602 in advance or storage is single In member 609.
The program for treating to be performed by computer can be for being held according to the order described in this manual with time sequencing The program of row processing, or can be for the program for parallel execution processing or necessary for example in the presence of execution processing when calling Program.
It should be noted that embodiment of the disclosure is not limited to above-described embodiment, the scope of the present disclosure can not departed from Under the premise of various modifications are carried out to above-described embodiment.
It is, for example, possible to use image of the super multi-view image generation in addition to refocusing image with capture (such as with scheming The image of the different brightness of the brightness of picture).
The parallax value of reference view and the parallax value of the occlusion area of reference view can be following value：The value is based on removing It is corresponding with each level to highest correlation parallax value average value outside highest correlation parallax value (such as highest phase The maximum of closing property parallax value).
In addition, imaging device 10 can not storage of hierarchically table.In this case, if only existing a kind of highest correlation Parallax value, then the highest correlation parallax value is arranged to the parallax value of reference view.On the other hand, if there is two kinds or more A variety of highest correlation parallax values, then by with reference view to highest correlation parallax value be arranged to regarding for reference view Difference, and the blocked area that reference view will be arranged in highest correlation parallax value in the highest correlation parallax value compared with distal side The parallax value in domain.
The capture images of correlation detection to be subjected to number be in the above-described embodiments 16, but can not be 16. Using according to the methods of Contrast Detection relevance values as correlation detection methods in the case of, three or more can be generated The relevance values of multiple capture images.
In addition, the imaging for the capture images that this technology can be applied not only to obtain many viewpoints by imaging unit is set It is standby, and can apply to obtain the imaging device of the capture images of many viewpoints by imaging len and microlens array.
The disclosure can also be embodied as structure described below.
(1) a kind of image processing equipment, including：
Computing unit, the computing unit calculate the capture figure of the multipair viewpoint using the capture images of multipair viewpoint The relevance values of picture；And
Generation unit, the generation unit generate reference view based on the relevance values calculated by the computing unit Different levels anaglyph, the anaglyph be the viewpoint anaglyph that is formed by the parallax value of the reference view with And anaglyph is blocked by what the parallax value of occlusion area was formed, the reference view is one of described multipair viewpoint.
(2) according to the image processing equipment described in (1), wherein, the generation unit be based on as with the relevance values In highest relevance values corresponding to the highest correlation parallax value of parallax value generate the viewpoint anaglyph and described Block anaglyph.
(3) according to the image processing equipment described in (2), wherein, the generation unit is based on the in the multipair viewpoint A pair of highest correlation parallax value generates the viewpoint anaglyph, and based on second pair of highest correlation parallax value Come generate it is described block anaglyph, described in the multipair viewpoint second pair with first pair of difference.
(4) according to the image processing equipment described in (3), wherein, the generation unit uses multiple described first pairs most Parallax value of the average value of high correlation parallax value as the viewpoint anaglyph, and using multiple described second pairs most The average value of high correlation parallax value blocks the parallax value of anaglyph as described in.
(5) image processing equipment according to (3) or (4), wherein, the generation unit is based on the correlation for making mutually to tackle The pattern of property value generates the viewpoint anaglyph and the screening with described first pair and second pair of associated table Keep off anaglyph.
(6) image processing equipment according to any one of (1) to (5), in addition to：
Virtual visual point image generation unit, the virtual visual point image generation unit are configured to be based on the multipair viewpoint Capture images and the anaglyph of the reference view that is generated by the generation unit generate the capture of virtual view Image.
(7) image processing equipment according to (6), in addition to：
Virtual focus point image generation unit, the virtual focus point image generation unit are configured to use the multipair viewpoint Capture images and generate virtual Jiao by the capture images of the virtual view of virtual visual point image generation unit generation The capture images of point.
(8) a kind of image processing method, including：
Calculation procedure, the correlation of the capture images of the multipair viewpoint is calculated using the capture images of multipair viewpoint Value；And
Generation step, the different levels of reference view are generated based on the relevance values calculated in the calculation procedure Anaglyph, the anaglyph is the viewpoint anaglyph that is formed by the parallax value of the reference view and by blocked area What the parallax value in domain was formed blocks anaglyph, and the reference view is one of described multipair viewpoint,
The calculation procedure and the generation step are performed by image processing equipment.
(9) a kind of image processing equipment, including：
Computing unit, the computing unit are configured to calculate the multipair viewpoint using the capture images of multipair viewpoint Capture images relevance values；And
Generation unit, the generation unit are configured to pattern based on the relevance values for making mutually to tackle and multipair regarded with described Predetermined in point generates viewpoint anaglyph to associated table, the viewpoint anaglyph according to as with described predetermined pair Highest relevance values corresponding to parallax value highest correlation parallax value and generate, the viewpoint anaglyph is by reference view Parallax value formed, the reference view is one of described multipair viewpoint.
(10) a kind of image processing method, including：
Calculation procedure, the correlation of the capture images of the multipair viewpoint is calculated using the capture images of multipair viewpoint Value；And
Generation step, the pattern based on the relevance values for making mutually to tackle make a reservation for associated with the multipair viewpoint Table generates viewpoint anaglyph, the viewpoint anaglyph according to as with it is described it is predetermined to highest relevance values it is corresponding The highest correlation parallax value of parallax value and generate, the viewpoint anaglyph is formed by the parallax value of reference view, the ginseng Viewpoint is examined as one of described multipair viewpoint,
The calculation procedure and the generation step are performed by image processing equipment.
Reference numerals list
10 imaging devices
13 graphics processing units
21 detection units
22 virtual visual point image generation units
23 refocusing image generation units
61-1 is to 61-6 correlation generation units
63 generation units

Claims (7)

1. a kind of image processing equipment, including：
Computing unit, the computing unit are configured to calculate catching for the multipair viewpoint using the capture images of multipair viewpoint Obtain the relevance values of image；And
Generation unit, the generation unit are configured to generate reference based on the relevance values calculated by the computing unit The anaglyph of the different levels of viewpoint, the anaglyph are the viewpoint disparity maps formed by the parallax value of the reference view As and by the parallax value of occlusion area is formed block anaglyph, the reference view is one of described multipair viewpoint.
2. image processing equipment according to claim 1, wherein, the generation unit be based on as with the relevance values In highest relevance values corresponding to the highest correlation parallax value of parallax value generate the viewpoint anaglyph and described Block anaglyph.
3. image processing equipment according to claim 2, wherein, the generation unit is based in the multipair viewpoint A pair of highest correlation parallax value generates the viewpoint anaglyph, and based on second pair of highest correlation parallax value Come generate it is described block anaglyph, described in the multipair viewpoint second pair with first pair of difference.
4. image processing equipment according to claim 3, wherein, the generation unit uses multiple described first pairs most Parallax value of the average value of high correlation parallax value as the viewpoint anaglyph, and using multiple described second pairs most The average value of high correlation parallax value blocks the parallax value of anaglyph as described in.
5. image processing equipment according to claim 1, in addition to：
Virtual visual point image generation unit, the virtual visual point image generation unit are configured to catching based on the multipair viewpoint The anaglyph of image and the reference view generated by the generation unit is obtained to generate the capture images of virtual view.
6. image processing equipment according to claim 5, in addition to：
Virtual focus point image generation unit, the virtual focus point image generation unit are configured to catching using the multipair viewpoint Obtain image and generate virtual focus point by the capture images of the virtual view of virtual visual point image generation unit generation Capture images.
7. a kind of image processing method, including：
Calculation procedure, the relevance values of the capture images of the multipair viewpoint are calculated using the capture images of multipair viewpoint；With And
Generation step, regarding for the different levels of reference view is generated based on the relevance values calculated in the calculation procedure Difference image, the anaglyph are the viewpoint anaglyphs that are formed by the parallax value of the reference view and by occlusion area What parallax value was formed blocks anaglyph, and the reference view is one of described multipair viewpoint,
The calculation procedure and the generation step are performed by image processing equipment.
CN201480030688.6A 2013-06-05 2014-05-23 Image processing equipment and image processing method Active CN105247860B (en)

Applications Claiming Priority (3)

Application Number Priority Date Filing Date Title
JP2013-118852 2013-06-05
JP2013118852 2013-06-05
PCT/JP2014/063646 WO2014196374A1 (en) 2013-06-05 2014-05-23 Image processing device and image processing method

Publications (2)

Publication Number Publication Date
CN105247860A CN105247860A (en) 2016-01-13
CN105247860B true CN105247860B (en) 2017-12-19

Family

ID=52008025

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201480030688.6A Active CN105247860B (en) 2013-06-05 2014-05-23 Image processing equipment and image processing method

Country Status (4)

Country Link
US (1) US10027947B2 (en)
JP (1) JP6551743B2 (en)
CN (1) CN105247860B (en)
WO (1) WO2014196374A1 (en)

Families Citing this family (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2016048402A2 (en) * 2014-09-26 2016-03-31 Reald Multiscopic image capture system
JP6536077B2 (en) * 2015-02-25 2019-07-03 沖電気工業株式会社 Virtual viewpoint image generating device and program
JP6631054B2 (en) * 2015-07-02 2020-01-15 沖電気工業株式会社 Virtual viewpoint image generation device and program
WO2017180615A1 (en) 2016-04-12 2017-10-19 Quidient, Llc Quotidian scene reconstruction engine
US10089788B2 (en) * 2016-05-25 2018-10-02 Google Llc Light-field viewpoint and pixel culling for a head mounted display device
WO2018014010A1 (en) 2016-07-15 2018-01-18 Light Field Lab, Inc. Selective propagation of energy in light field and holographic waveguide arrays
JP7030452B2 (en) * 2017-08-30 2022-03-07 キヤノン株式会社 Information processing equipment, information processing device control methods, information processing systems and programs
US11650354B2 (en) 2018-01-14 2023-05-16 Light Field Lab, Inc. Systems and methods for rendering data from a 3D environment
US10967565B2 (en) 2018-01-14 2021-04-06 Light Field Lab, Inc. Energy field three-dimensional printing system
AU2019262147A1 (en) 2018-05-02 2020-12-10 Quidient, Llc A codec for processing scenes of almost unlimited detail
EP3598390A1 (en) * 2018-07-19 2020-01-22 Thomson Licensing Method for estimating a depth for pixels, corresponding device and computer program product
JP2023034160A (en) * 2021-08-30 2023-03-13 キヤノン株式会社 Accuracy estimation device, imaging device, accuracy estimation method, control method and program

Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101822068A (en) * 2007-10-11 2010-09-01 皇家飞利浦电子股份有限公司 Method and device for processing depth-map
CN101873507A (en) * 2009-04-22 2010-10-27 索尼公司 Transmitting device, receiving system, stereoscopic image data transmission method and method of reseptance
CN101516040B (en) * 2008-02-20 2011-07-06 华为终端有限公司 Video matching method, device and system

Family Cites Families (24)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CA2259411A1 (en) * 1997-05-05 1998-11-12 Shell Oil Company Visual recognition method
JP3593466B2 (en) 1999-01-21 2004-11-24 日本電信電話株式会社 Method and apparatus for generating virtual viewpoint image
JP2006173038A (en) * 2004-12-20 2006-06-29 Hitachi High-Technologies Corp Charged particle beam device, sample image display method, and image shift sensitivity measuring method
CN100584013C (en) * 2005-01-07 2010-01-20 日本电信电话株式会社 Video encoding method and device, video decoding method and device
KR100968920B1 (en) * 2006-01-05 2010-07-14 니폰덴신뎅와 가부시키가이샤 Video encoding method, decoding method, device thereof, program thereof, and storage medium contains the program
WO2008035665A1 (en) * 2006-09-20 2008-03-27 Nippon Telegraph And Telephone Corporation Image encoding method, decoding method, device thereof, image decoding device, program thereof, and storage medium containing the program
CN100576934C (en) * 2008-07-03 2009-12-30 浙江大学 Virtual visual point synthesizing method based on the degree of depth and block information
PL2342693T3 (en) * 2008-09-25 2012-12-31 Koninl Philips Electronics Nv Three dimensional image data processing
JP4706882B2 (en) 2009-02-05 2011-06-22 ソニー株式会社 Imaging device
KR101940023B1 (en) * 2009-02-19 2019-01-21 톰슨 라이센싱 3d video formats
JP5504881B2 (en) 2009-12-25 2014-05-28 ソニー株式会社 Arithmetic apparatus, arithmetic method, arithmetic program, and microscope
JP5287702B2 (en) * 2009-12-25 2013-09-11 ソニー株式会社 Image processing apparatus and method, and program
JP5387856B2 (en) * 2010-02-16 2014-01-15 ソニー株式会社 Image processing apparatus, image processing method, image processing program, and imaging apparatus
GB2478157A (en) * 2010-02-26 2011-08-31 Sony Corp Method and apparatus for cutting between a first and second image sequence in a stereoscopic video
CN101808251B (en) * 2010-04-06 2011-07-20 浙江大学 Method for extracting blocking information in stereo image pair
TW201200959A (en) * 2010-06-29 2012-01-01 Fujifilm Corp One-eyed stereo photographic device
CN103189715B (en) * 2010-11-05 2015-12-02 松下知识产权经营株式会社 Stereoscopic image processing device and stereoscopic image processing method
US20120113235A1 (en) * 2010-11-08 2012-05-10 Sony Corporation 3d glasses, systems, and methods for optimized viewing of 3d video content
US8878950B2 (en) * 2010-12-14 2014-11-04 Pelican Imaging Corporation Systems and methods for synthesizing high resolution images using super-resolution processes
US9055284B2 (en) * 2010-12-20 2015-06-09 Panasonic Intellectual Property Management Co., Ltd. Stereo image processing apparatus and stereo image processing method
CN103404156A (en) * 2011-02-24 2013-11-20 索尼公司 Image processing apparatus and image processing method
WO2012157210A1 (en) * 2011-05-19 2012-11-22 パナソニック株式会社 Three-dimensional imaging device, image processing device, image processing method, and program
JP2013026826A (en) * 2011-07-21 2013-02-04 Sony Corp Image processing method, image processing device and display device
JP5868055B2 (en) 2011-07-26 2016-02-24 キヤノン株式会社 Image processing apparatus and image processing method

Patent Citations (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN101822068A (en) * 2007-10-11 2010-09-01 皇家飞利浦电子股份有限公司 Method and device for processing depth-map
CN101516040B (en) * 2008-02-20 2011-07-06 华为终端有限公司 Video matching method, device and system
CN101873507A (en) * 2009-04-22 2010-10-27 索尼公司 Transmitting device, receiving system, stereoscopic image data transmission method and method of reseptance

Also Published As

Publication number Publication date
US10027947B2 (en) 2018-07-17
CN105247860A (en) 2016-01-13
US20160105660A1 (en) 2016-04-14
WO2014196374A1 (en) 2014-12-11
JPWO2014196374A1 (en) 2017-02-23
JP6551743B2 (en) 2019-07-31

Similar Documents

Publication Publication Date Title
CN105247860B (en) Image processing equipment and image processing method
CN101124815B (en) Imaging device, imaging method, and imaging device design method
CN102883176B (en) Image processing apparatus, image processing method
CN102812715A (en) Three-dimensional imaging device and three-dimensional imaging method
WO2018163843A1 (en) Imaging device and imaging method, and image processing device and image processing method
US10074343B2 (en) Three-dimensional image output apparatus and three-dimensional image output method
CN103279952A (en) Target tracking method and device
WO2019033673A1 (en) Panoramic sea view monitoring method and device, server and system
US20120105601A1 (en) Apparatus and method for creating three-dimensional panoramic image by using single camera
Jin et al. Occlusion-aware unsupervised learning of depth from 4-d light fields
Jin et al. Light field super-resolution via attention-guided fusion of hybrid lenses
US20100253682A1 (en) Image generating apparatus and computer program
US8731326B2 (en) Object recognition system and method
Wu et al. Novel image copy detection with rotating tolerance
Zhou et al. FANet: Feature aggregation network for RGBD saliency detection
Shin et al. Super-resolution method based on multiple multi-layer perceptrons for iris recognition
Zhou et al. Hierarchical visual comfort assessment for stereoscopic image retargeting
Kang et al. Camera model identification using a deep network and a reduced edge dataset
Chen et al. Depth map generation based on depth from focus
EP2438572A1 (en) Generating images with different fields of view
Islam et al. Robust enhancement of depth images from depth sensors
Chen et al. FAFNet: Fully aligned fusion network for RGBD semantic segmentation based on hierarchical semantic flows
CN109978928A (en) A kind of binocular vision solid matching method and its system based on Nearest Neighbor with Weighted Voting
CN105301863B (en) Liquid crystal lens imaging device and liquid crystal lens imaging method
Zhu et al. DANet: dynamic salient object detection networks leveraging auxiliary information

Legal Events

Date Code Title Description
C06 Publication
PB01 Publication
C10 Entry into substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant